Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

FIX: Decision Tree Visualization #386

Merged
merged 12 commits into from
Aug 1, 2023
Merged

FIX: Decision Tree Visualization #386

Show file tree
Hide file tree
Changes from 2 commits
Commits
Show all changes
12 commits
Select commit Hold shift + click to select a range
4b87c84
Fix DecisionTree Visualization
reidjohnson Jul 24, 2023
1b0a86a
Update visualization tests
reidjohnson Jul 25, 2023
52e5efa
Visualize type object
reidjohnson Jul 26, 2023
1aec5e9
Add visualize tests for decision trees
reidjohnson Jul 26, 2023
3cbf6e5
Revert changes
reidjohnson Jul 26, 2023
a16bc7c
Pass trust info to child
reidjohnson Jul 27, 2023
369c206
Pass as explicit kwargs and update docstring
reidjohnson Jul 27, 2023
797bf1a
Add changelog
reidjohnson Jul 27, 2023
316da5c
Wrap constructor in TypeNode
reidjohnson Jul 28, 2023
328cd52
Update DT unit test names
reidjohnson Jul 28, 2023
5d55e61
Use get_module
reidjohnson Jul 30, 2023
aa52598
Fix DT unit test for sklearn 1.4
reidjohnson Aug 1, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
10 changes: 9 additions & 1 deletion skops/io/_visualize.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -232,7 +232,15 @@ def walk_tree(
"here: https://github.com/skops-dev/skops/issues"
)

if node_name == "constructor":
if isinstance(node, type):
yield NodeInfo(
level=level,
key=node_name,
val=type(node).__name__,
is_self_safe=False,
is_safe=False,
Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

here we should defer to the information in the Node instead of assuming it to be unsafe all the time. the Tree constructor for instance, is safe.

Copy link
Contributor Author

@reidjohnson reidjohnson Jul 27, 2023 •
edited
Loading

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Right, I agree. However the challenge I'm running into is that it's not a valid Node object, so it does not have the necessary attributes:

AttributeError: type object 'sklearn.tree._tree.Tree' has no attribute 'is_self_safe'

If I understand correctly, it's because the constructor class is directly set as a child node here.

Copy link
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

so probably the parent when doing the walk_tree should send this info to the child.

is_last=is_last,
)
return

if isinstance(node, dict):
Expand Down
58 changes: 56 additions & 2 deletions skops/io/tests/test_visualize.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,14 +4,15 @@

import pytest
import sklearn
from sklearn.ensemble import RandomForestRegressor
from sklearn.linear_model import LogisticRegression
from sklearn.pipeline import FeatureUnion, Pipeline
from sklearn.preprocessing import (
FunctionTransformer,
MinMaxScaler,
PolynomialFeatures,
StandardScaler,
)
from sklearn.tree import DecisionTreeClassifier, DecisionTreeRegressor

import skops.io as sio

Expand Down Expand Up @@ -39,7 +40,7 @@ def unsafe_function(x):
("scale", MinMaxScaler()),
])),
])),
("clf", RandomForestRegressor(random_state=0)),
("clf", LogisticRegression(random_state=0, solver="lbfgs")),
]).fit([[0, 1], [2, 3], [4, 5]], [0, 1, 2])
# fmt: on
return pipeline
Expand Down Expand Up @@ -282,3 +283,56 @@ def test_long_bytes(self, capsys):
]
stdout, _ = capsys.readouterr()
assert stdout.strip() == "\n".join(expected)

@pytest.mark.parametrize("cls", [DecisionTreeClassifier, DecisionTreeRegressor])
def test_decision_tree(self, cls, capsys):
model = cls(random_state=0).fit([[0, 1], [2, 3], [4, 5]], [0, 1, 2])
dumped = sio.dumps(model)
sio.visualize(dumped)

classes = []
if isinstance(model, DecisionTreeClassifier):
criterion = "gini"
classes = [
" ├── classes_: numpy.ndarray",
" ├── n_classes_: numpy.int64",
]
elif isinstance(model, DecisionTreeRegressor):
criterion = "squared_error"

expected = [
"root: sklearn.tree._classes.{}".format(cls.__name__),
"└── attrs: builtins.dict",
' ├── criterion: json-type("{}")'.format(criterion),
' ├── splitter: json-type("best")',
" ├── max_depth: json-type(null)",
" ├── min_samples_split: json-type(2)",
" ├── min_samples_leaf: json-type(1)",
" ├── min_weight_fraction_leaf: json-type(0.0)",
" ├── max_features: json-type(null)",
" ├── max_leaf_nodes: json-type(null)",
" ├── random_state: json-type(0)",
" ├── min_impurity_decrease: json-type(0.0)",
" ├── class_weight: json-type(null)",
" ├── ccp_alpha: json-type(0.0)",
" ├── n_features_in_: json-type(2)",
" ├── n_outputs_: json-type(1)",
]
expected += classes
expected += [
" ├── max_features_: json-type(2)",
" ├── tree_: sklearn.tree._tree.Tree",
" │ ├── attrs: builtins.dict",
" │ │ ├── max_depth: json-type(2)",
" │ │ ├── node_count: json-type(5)",
" │ │ ├── nodes: numpy.ndarray",
" │ │ └── values: numpy.ndarray",
" │ ├── args: builtins.tuple",
" │ │ ├── content: json-type(2)",
" │ │ ├── content: numpy.ndarray",
" │ │ └── content: json-type(1)",
" │ └── constructor: type [UNSAFE]",
' └── _sklearn_version: json-type("{}")'.format(sklearn.__version__),
]
stdout, _ = capsys.readouterr()
assert stdout.strip() == "\n".join(expected)